G. Multiple images back | top | MacLIspix Home
This example shows some of the tools in MacLispix that are designed to handle data sets consisting of several images. The data that are in the MLx Demo:Super Cond Phases folder consists of a set of quantitated x-ray maps (Bernik, 94) of some ceramic superconductor material.
Erase image windows and arrays using the MLx -> Pixel Array Ops - Erase Everything command. This might not be necessary, depending on how much memory you have assigned to MacLispix, but it might be a good idea in order to avoid screen clutter.
Set the image default to the Super Cond Phases folder (if desired) by using the File -> Open MLx -> Demo command to open the 'about supcon phases.doc file. Then use command-D, and the DBMac -> Defaults -> Set Image Dir command, as before.
Load a set of x-ray maps using the MLx -> Image Files -> EPM -> Read command, in the MLx Demo:Super Cond Phases:Sample Data:Set 10 1 folder. This example shows how this command is designed to read groups of images: select one of the images as an example, and then press the 'Select All' button in the next dialog. All of the images will be loaded into memory, as shown in the Listener or by using the MLx -> Pixel array Ops -> Info command. The EPM (for Electron Probe Map) format is a special one, used by the Microanalysis group for quantitated maps. The format consists of a 128 byte header, and then 16 bit integer pixel values, where the value is the weight percent of the element times 100 (or times 10 for the Total maps). These data are loaded in a different way by the demonstration software where the arrays are real rather than integer, and represent the mole fraction of the elements.
Display three of the images using the MLx -> Show Image Arrays command.
Select the Bi, Cu and Sr maps. The element symbols are embedded in the file
If auto_display is turned on, all of the maps will be displayed at this point. The illustrations in the paper apply to having only three of the maps displayed, but will work with minor adjustments if all of the images are displayed.
Zoom them x2 with the MLx -> Image Windows -> Zoom command using 2 as the zoom factor, and selecting them or, if no other images are displayed, pressing the 'Select All' in the dialog that follows. Next, tile them with the MLx -> Image Windows Tile command, to see them all at once. The DBMac - For Slides -> Redraw Windows command can be used to fill in the grow squares in the lower right corners of the maps.
Inspect individual pixel values using the MLx -> Show Values -> Read Single Pixel command. The data window will appear in the lower right corner of the screen once the mouse is clicked on one of the images. Further mouse clicks on any of the images that are listed in the windoid to the lower right (the arrays have the same dimensions as that of the window first clicked) will cause the pixel coordinates and values in all of the images to be displayed.
Sometimes the correspondence of the images can be examined by drawing
on one of them using the MLx -> Graphics -> Draw on Images command,
and causing the drawing to be mimicked on the other images. Click the 'Select
All' button to draw on and see the results on all of the images. For purposes
of an example, zoom two of the images by dragging the lower right corners
with the mouse, and repositione them so that they can all be seen by dragging
on their title bars. (The images can be made smaller as well as larger.)
Draw the outlines with the mouse on any of the images at any time. The lines
will be 'pantogrammed' to the other images. Any of the images can be used
to do the drawing, but the lines will be smoother the larger the image is,
since small windows have fewer positions for the width of the window, which
is 50 µm on the sample. Use the general LISP interrupt, command-.
(command-period) to stop drawing.
Spatial correspondence of the images can also be observed at a glance
(for two or three images) using a color overlay. Invoke the MLx -> Image
Windows -> RGB Overlay -> Pick RGB Images command, and choose the
Bi map for the red image, the Cu map for the green, and Sr for the blue.
Zoom the image by dragging the grow square in the lower right. Note that
the various phases of this material correspond to different pastel shades
-- the higher the percentage of copper, the more green the color. The superconducting
phase here appears pink -> it is predominantly red (a large proportion
of Bismuth), less blue (less Strontium), and a little green (some Copper).
This display is quick, but qualitative. The colors appear somewhat blotchy
when the Monitors Control Panel is set for 256 colors. Thousands or millions
of colors are recommended. Of course, if the panel has been set for 256
shades of gray for previous steps in this tutorial, the color overlay will
look like a black and white image.
A more quantitative display of this same information is shown in the scatter diagrams, or "Concentration Histogram Images" (see Bright, 1991 or Bright, 1992 for color renderings of similar displays).
To make a scatter diagram corresponding to the previous color overlay,
use the MLx -> C Hist Image -> Setup -> RGB->XYZ-> Label
command to quickly choose the images for each axis and label the axes. Then
invoke the MLx - C Hist Image -> 3D normal command, and when it is finished,
adjust the thermal scale by moving the right edge of the yellow slider bar
with the mouse as desired. A value of 200 for the right edge gives a satisfactory
thermal color display of the scatter diagram.
Which cluster corresponds to which area of the maps may be visualized
by using the MLx -> C Hist Image -> 3D traceback command. Make sure
the scatter diagram window is in front first. Also make sure that the bottom
of the Listener window says 'idle' when it is in front (click on it). Use
the default name for the traceback window, and follow the directions in
the dialog that follows: use the mouse to enclose the cluster of interest,
then hold down the control key and click inside the enclosed area to fill
it in. Press the command key when the filled in areas are satisfactory.
The traceback window is in register with the original maps (images), ane
the while pixels correspond to histogram pixels of the filled in areas.
The white TrBk 1 pixels can be outlined on the other images with the MLx -> Masks -> Outline Using Mask command, selecting TrBk 2 as the mask window, and by selecting (with the command key pressed) the three maps and the color overlay. (Sometimes the outline color fills in one of the original images. This is a bug. Just try the command over again.
The three dimensional histogram is represented here as a projection.
The point of view for the projection can be changed and tested with the
MLx - C Hist Image -> Setup -> Rotation and -> tilt commands. A
stereo pair may be made by doing the histogram over again after changing
the rotation by a few degrees.
The two dimensional histogram can be shown using the x and y images as they are at this point. The histogram will have scales on its axes that correspond to the range of the original data (the histogram itself is made using the scaled data -> which allows for flexibility of scaling, and affects the appearance of the histogram only slightly if at all). Use the MLx -> C Hist Image (CCH) - 2D (CCH) command.
Other treatment of this data is described in: (edx "CCL:MLx Demo;Super Cond Phases;'about supcon phases.doc").
Other demonstrations are described in: (edx "CCL:MLx Demo;Demo Instructions.txt")